Novel game playing strategies using adaptive data structures

Abstract

The problem of achieving intelligent game play has been a historically important, widely studied problem within Artificial Intelligence (AI). Although a substantial volume of research has been published, it remains a very active research area with new, innovative techniques introduced regularly. My thesis work operates on the hypothesis that the natural ranking mechanisms provided by the previously unrelated field of Adaptive Data Structures (ADSs) may be able to improve existing game playing strategies. Based on this reasoning, I have examined the applicability of ADSs in a wide range of areas within game playing, leading to the creation of two novel techniques, the Threat-ADS and History-ADS. I have found that ADSs can produce substantial improvements under a broad range of game models and configurations.